TGlow: repository for image-analysis instances and analysis scripts

This repository contains instance directories and analysis scripts used to support the TGlow manuscript and analysis (preprint linked below). It includes example instances, data-analysis pipelines, and helper scripts to reproduce figures and results from the study.

The preprint presents TGlow, a reproducible computational workflow and R package for processing, normalizing, and analyzing high-content cell-imaging data to study T cell activation and perturbation responses. The pipelines standardize image-derived feature extraction, quality control, scaling, and regression analyses to facilitate comparative studies across experimental instances.

Absolute paths have been replaced with:

• <REPO_ROOT> for the root of this repo
• <PIPELINE_ROOT> for the root of the Nextflow pipeline installation folder
• <TGLOW_LIB> for the location of the library where the tglow-r package is installed. The lib='<TGLOW_LIB>' can be removed in case you installed in the default enviroment. We only install this in a seperate library to enable easy runtime re-installs when using Rstudio server.
• <RAW_ROOT> for the path to raw Harmony exports (this one can be ignored)

If reproducing these need to be updated, which is easy to do in vscode with a global find replace over all files

Contents

• CRISPR_1: CRISPR data instance
• DRUG_PERTURB: Drug perturbation data instance
• MUNCHKIN: CD4 activation timecourse instance
• TIMECOURSE_CD8_EXHAUSTION_PILOT: CD8 Exhaustion instance

Each feature_analysis directory contains R analysis scripts, input parameter files, and output folders (objects, reductions, regressions, QC). Each pipeline directory contains the image-processing pipeline scripts and inputs used to generate the analysis-ready data.

Quick links

• TGlow pipeline implementation: https://github.com/TrynkaLab/tglow-pipeline
• TGlow R analysis package: https://github.com/TrynkaLab/tglow-r
• Manuscript (preprint): https://www.biorxiv.org/content/10.64898/2026.02.10.704860v1
• TGlow objects supporting this analysis: https://www.ebi.ac.uk/biostudies/studies/S-BSST2653

Reproducing analyses in this repo

1. Inspect the instance of interest (for example, CRISPR_1/feature_analysis/).
2. Review parameter files in each instance's input (e.g. feature_analysis/input/) and adjust paths as needed.
3. Fetch the data from image archive (see manuscript) and place it in <INSTANCE>/pipeline/results/images
4. Run the pipeline scripts in the instance's pipeline/scripts/ to recreate image-derived objects.
5. Run the R analysis scripts in feature_analysis/scripts/ to reproduce QC, scaling, regressions, and figures.

Dependencies

• R & tglow-r (version used for analyses documented in each instance's scripts)
• Nextflow >= 25.04.6 & tglow-pipeline

How to cite Please cite the preprint: https://www.biorxiv.org/content/10.64898/2026.02.10.704860v1 and the TGlow repositories linked above when using these scripts or reproducing results.